loopx-agent · Nov 4, 2026 · Nov 2, 2026 · Nov 4, 2026
@@ -0,0 +1,110 @@
+# Global Todo readiness owner cutover — 2026-11-02
+
+## Boundary and economics
+
+Baseline: `fb16c6cb1369efce52a673030c914b38e1dac166`. This bounded slice
+moves the complete Global Todo readiness decision from `global_todos.py` to
+`projectGlobalTodoReadiness` in the existing `todos/decision_scope.ts` owner,
+registered as `todo.global_readiness.project`. Both `loopx global-todos` and
+`loopx goal-portfolio` use it through their shared display adapter; portfolio
+classifies each registered Agent independently.
+
+TypeScript owns ordered selected/lane/summary/frontier traversal, first-source
+merge with missing-field fill, source/status provenance, in-process `todoGateRelation`, blocked/deferred/runnable
+precedence, and warning ordering. `normal_delivery_allowed` remains a readiness
+projection signal and grants no execution permission. Existing readiness values
+are reused; the new versioned request/result describe this bounded projection,
+not a new lifecycle or authority vocabulary.
+
+Instrumentation pins all cross-runtime requests, including envelope sealing:
+
+| Operation | Baseline | Candidate |
+| --- | --- | --- |
+| One classification, gates and candidates present | 1 relation request | 1 complete readiness request |
+| One classification, no gates | 0 | 1 |
+| One classification, no candidates (including empty payload) | 0 | 1 |
+| Global command, one classified Goal with gates | 2 including envelope | 2 including envelope |
+| Global command, gate-free or empty classification | 1 envelope | 2 including envelope |
+| Portfolio, two Agents with gates | 2 | 2 |
+| Portfolio, two gate-free/empty Agent classifications | 0 | 2 |
+
+The gate-free/empty increase is intentional and disclosed. No latency or
+long-horizon effectiveness/efficiency benefit is claimed.
+
+Product physical diff: +163/−166 lines, net −3. Python removes source
+traversal/merge helpers, blocked/deferred classification, readiness precedence
+composition and its separate safety-rank table. The uncalled Python scalar
+`todo_gate_relation` bridge and its TS scalar dispatch are deleted. Its synthetic
+smoke and decoder tests use the production batched bridge retained for summary.
+
+Python retains `open_user_gate_todo_items` as the single gate-selection owner:
+summary, stall repair, Todo summary and agent scope still call it. Its ordered
+selection, done filtering and Python-equality deduplication must not be copied
+into TypeScript. Selected gates cross as decoded facts with target text.
+Shared `_facts` retains persisted ID/scope normalization and legacy task-class
+interpretation. The bounded request carries only selected/lane records, delivery
+signal, and list-valued summary/frontier fields. Python decodes dict elements of
+every such list generically; TypeScript alone owns candidate list-key traversal.
+There is no recursive payload walk. The largest frozen case's compact UTF-8 JSON
+request shrinks from 9,749 to 6,447 bytes (33.9%); outputs remain byte-identical.
+Python also retains status/quota IO, public text redaction, `_work_kind` display
+interpretation, final grouping/deduplication/display sorting, envelope sealing
+and Markdown. These compatibility and presentation boundaries are not a second
+candidate-readiness owner.
+
+Facade exit: delete `project_global_todo_readiness` and its decoded-field
+sidecar once both global and portfolio host/read models consume the typed Todo
+snapshot and projection directly, with native legacy metadata decoding and
+public-safe rendering at their owning boundaries. Delete the shared display
+adapter when neither consumer needs Python normalization. Retire the temporary
+characterization corpus/archive comparison at that exit after durable semantic
+and public-entrypoint tests cover the retained contracts.
+
+## Qualification and limits
+
+Seven synthetic characterization cases passed on unchanged Python before the
+cutover; their complete JSON/Markdown bytes and baseline/candidate crossing
+lists were then frozen. The fixture SHA-256 is
+`eaf17921f81357329ba7bc3c7dcb34253eed83b53d31c31d6dba0fdb5c547641`.
+Cases cover complementary duplicate fields, conflicting statuses, delivery
+denial, exact/scope conflict, unrelated gates, blocked/deferred overlap,
+missing identities, two Agents, warning order, empty inputs and public-text
+redaction. Independent TS tests assert precedence, denial and exact/scope
+conflict semantics, plus malformed protocol rejection. Python decoder tests
+reject malformed results without fallback.
+
+`tests/test_global_todo_readiness_characterization.py::test_production_cli_archive_parity`
+archives `HEAD` without stash and compares both public commands, JSON and
+Markdown, on the same synthetic temporary home with real File and SQLite
+providers. Only clocks are frozen before importing the production CLI; field
+presence, diagnostics, source versions and hashes are compared unchanged.
+All eight command/provider/format pairs match and require populated Todo and
+Agent results. This qualifies a source checkout, not installed packaging or
+frontend behavior; no caller capability or user journey was added.
+
+The focused scheduler-state readback test
+`test_real_cli_quota_should_run_discloses_unreadable_scheduler_state` fails on
+both the changed checkout and the archived baseline: its quota command omits
+Turn identity and returns `QUOTA_VALIDATION_FAILED` with the requirement for
+`--turn-instance-id` or `--begin-turn`. This is a pre-existing caller/validation
+mismatch, outside readiness ownership. Preserve that failure; do not relax the
+argument gate or rewrite this migration's expected outputs to conceal it.
+
+Semantic drift remains `independently_maintained=45/45`; there are no new
+control-plane source files or same-basename pairs. Maintainability and interface
+budgets pass unchanged. The integrated canary passes File and SQLite. The
+diff-scoped advisory detects no supported new vocabulary carriers across four
+changed product source files; its supported scope excludes test sources.
+
+## Review correction validation
+
+After removing duplicate gate selection and bounding the request: requested
+Python suite 151 passed / 1 failed (the scheduler caller mismatch above);
+native Node tests 19 passed / 0 failed; boundary suite 28 passed / 0 failed.
+Ruff on all four changed Python files, control-plane typecheck, both requested
+smokes and diff whitespace checks pass. Seven frozen characterization cases
+remain byte-identical, with the fixture hash unchanged. File/SQLite production
+CLI parity covers all eight command/format/provider pairs. A focused boundary
+regression proves generic decoding of unknown list fields, omission of unrelated
+objects and reuse of shared gate deduplication. No commit, staging or push was
+performed; generated uv.lock was removed.
@@ -0,0 +1,91 @@
+# Global Todo 就绪分类归属切换 — 2026-11-02
+
+## 边界与经济性
+
+基线：`fb16c6cb1369efce52a673030c914b38e1dac166`。本次有界切片将
+完整 Global Todo 就绪决策从 `global_todos.py` 移至现有
+`todos/decision_scope.ts` owner 的 `projectGlobalTodoReadiness`，注册为
+`todo.global_readiness.project`。`loopx global-todos` 与
+`loopx goal-portfolio` 均通过共享展示适配器使用它；portfolio 对每个已注册
+Agent 分别分类。
+
+TypeScript 负责 selected/lane/summary/frontier 的有序遍历、首个来源优先并补齐
+缺失字段的合并、来源/状态溯源、进程内
+`todoGateRelation`、blocked/deferred/runnable 优先级，以及警告顺序。
+`normal_delivery_allowed` 仍只是就绪投影信号，不授予执行权限。复用现有就绪值；
+新增版本化请求/结果仅描述本次有界投影，不构成新的生命周期或权限词汇。
+
+测试插桩固定所有跨运行时请求，包括 envelope sealing：
+
+| 操作 | 基线 | 候选 |
+| --- | --- | --- |
+| 单次分类，同时存在 gate 与候选 | 1 次关系请求 | 1 次完整就绪请求 |
+| 单次分类，无 gate | 0 | 1 |
+| 单次分类，无候选（含空 payload） | 0 | 1 |
+| Global 命令，分类一个有 gate 的 Goal | 2，含 envelope | 2，含 envelope |
+| Global 命令，无 gate 或空分类 | 1 次 envelope | 2，含 envelope |
+| Portfolio，两个有 gate 的 Agent | 2 | 2 |
+| Portfolio，两个无 gate/空分类的 Agent | 0 | 2 |
+
+无 gate/空路径的增加是有意且已披露的。不声称延迟或长程效果/效率收益。
+
+产品物理行差异：+163/−166 行，净 −3。Python 删除来源遍历/合并 helper、
+blocked/deferred 分类、就绪优先级组合及独立 safety-rank 表。删除无生产调用方的
+Python 标量 `todo_gate_relation` 桥及其 TS dispatch；对应 smoke 和 decoder 测试
+使用 summary 仍在调用的批量关系桥。
+
+Python 保留 `open_user_gate_todo_items` 为唯一 gate 选择 owner：summary、
+stall repair、Todo summary 和 agent scope 仍依赖它。有序选择、done 过滤及
+Python 相等去重不可复制到 TypeScript。已选 gate 以解码 facts 和目标文本传输。
+共享 `_facts` 保留持久化 ID/scope 归一化与旧 task-class 解释。有界请求仅传输
+selected/lane 记录、delivery 信号以及 summary/frontier 的 list 字段。Python
+通用解码每个 list 的 dict 元素；TypeScript 独占候选 list key 遍历规则，不再递归
+遍历整个 payload。最大冻结用例的紧凑 UTF-8 JSON 请求从 9,749 降至 6,447 字节
+（减少 33.9%），输出字节不变。Python 还保留 status/quota IO、公开文本脱敏、
+`_work_kind` 展示解释、最终分组/去重/展示排序、envelope sealing 和 Markdown。
+这些兼容与展示边界不是第二个候选就绪 owner。
+
+Facade 退出条件：当 global 与 portfolio 的宿主/读模型都直接消费 typed Todo
+snapshot 和投影，且各自边界已原生承接旧元数据解码及公开安全渲染时，删除
+`project_global_todo_readiness` 及其字段解码 sidecar。当两个消费者均不再需要
+Python 归一化时，删除共享展示适配器。在该退出点，持久语义与公开入口测试覆盖
+保留契约后，移除临时 characterization corpus/archive 对照。
+
+## 资格与限制
+
+七个合成 characterization 用例先在未改动的 Python 上通过，然后冻结完整
+JSON/Markdown 字节及基线/候选 crossing 列表。fixture SHA-256 为
+`eaf17921f81357329ba7bc3c7dcb34253eed83b53d31c31d6dba0fdb5c547641`。
+覆盖互补重复字段、冲突状态、delivery 拒绝、精确目标/scope 冲突、无关 gate、
+blocked/deferred 重叠、缺失身份、两个 Agent、警告顺序、空输入与公开文本脱敏。
+独立 TS 测试断言优先级、拒绝及精确目标/scope 冲突语义，并拒绝畸形协议。
+Python decoder 测试拒绝畸形结果且不作回退。
+
+`tests/test_global_todo_readiness_characterization.py::test_production_cli_archive_parity`
+通过 archive `HEAD`、不用 stash，在同一合成临时 home 上以真实 File 和 SQLite
+provider 对比两个公开命令的 JSON 与 Markdown。仅在导入生产 CLI 前冻结时钟；
+字段存在性、诊断、来源版本与哈希均原样比较。八组命令/provider/格式均一致，
+并要求结果实际包含 Todo 与 Agent。此证据资格为 source checkout，不覆盖安装包
+或前端行为；没有新增调用方能力或用户旅程。
+
+聚焦 scheduler-state 读回测试
+`test_real_cli_quota_should_run_discloses_unreadable_scheduler_state` 在修改树与
+archive 基线上都失败：其 quota 命令缺少 Turn 身份，返回
+`QUOTA_VALIDATION_FAILED`，要求 `--turn-instance-id` 或 `--begin-turn`。
+这是已有的调用方/验证不匹配，位于就绪 owner 之外。保留该失败；不得放宽参数门禁，
+也不得改写本次迁移的期望输出来掩盖它。
+
+语义漂移仍为 `independently_maintained=45/45`；未新增 control-plane 源文件或
+同名配对。Maintainability 与接口预算保持不变并通过。集成 canary 的 File 和
+SQLite 均通过。diff 范围 advisory 在四个变更产品源文件中未发现受支持的新词汇
+载体；该工具的支持范围不包含测试源文件。
+
+## 评审修正验证
+
+删除重复 gate 选择并收窄请求后：指定 Python 套件 151 通过 / 1 失败（上述
+scheduler 调用参数不匹配）；原生 Node 测试 19 通过 / 0 失败；边界套件
+28 通过 / 0 失败。四个已修改 Python 文件的 Ruff、control-plane typecheck、
+两个指定 smoke 和 diff 空白检查均通过。七个冻结 characterization 用例的输出
+字节不变，fixture 哈希不变。File/SQLite 生产 CLI 对照覆盖八组命令/格式/provider。
+聚焦边界回归验证未知 list 字段的通用解码、排除无关对象及复用共享 gate 去重。
+未 commit、stage 或 push；已删除生成的 uv.lock。
@@ -2845,3 +2845,4 @@ with shared write/read status-phase checks. See the [2026-10-02 checkpoint](ledg
 - Complete durable-completion evidence read: [2026-11-02 owner and crossings ledger](ledger/typescript-control-plane-migration-v0/2026-11-02-completion-evidence-read.md).
 - Complete adaptive child admission decision: [2026-11-02 checkpoint](ledger/typescript-control-plane-migration-v0/2026-11-02-adaptive-child-admission.md).
 - Progress-review receipt qualification: [2026-11-02 checkpoint](ledger/typescript-control-plane-migration-v0/2026-11-02-progress-review-qualification.md).
+- Global Todo readiness complete owner cutover (both consumers): [2026-11-02 ledger](ledger/typescript-control-plane-migration-v0/2026-11-02-global-todo-readiness.md).
@@ -2110,3 +2110,4 @@ SQLite 中位数略升，收益不能外推为全面加速。这只资格化本
 - 完整 durable-completion 证据读取：[2026-11-02 owner 与调用边界记录](ledger/typescript-control-plane-migration-v0/2026-11-02-completion-evidence-read.zh-CN.md)。
 - 完整自适应子 Agent 准入决策：[2026-11-02 检查点](ledger/typescript-control-plane-migration-v0/2026-11-02-adaptive-child-admission.zh-CN.md)。
 - 进度评审 receipt 资格判定：[2026-11-02 检查点](ledger/typescript-control-plane-migration-v0/2026-11-02-progress-review-qualification.zh-CN.md)。
+- Global Todo 就绪分类完整 owner 切换（两个消费者）：[2026-11-02 记录](ledger/typescript-control-plane-migration-v0/2026-11-02-global-todo-readiness.zh-CN.md)。
@@ -11,7 +11,7 @@
 if str(REPO_ROOT) not in sys.path:
     sys.path.insert(0, str(REPO_ROOT))
 
-from loopx.control_plane.todos.decision_scope import todo_gate_relation  # noqa: E402
+from loopx.control_plane.todos.decision_scope import todo_gate_relations  # noqa: E402
 from loopx.control_plane.scheduler.execution_context import (  # noqa: E402
     GENERIC_CLI_OUTER_CONTROLLER_SCHEDULER_CONTEXT,
     scheduler_execution_context_for_runtime_profile,
@@ -967,7 +967,7 @@ def assert_conflicting_exact_and_decision_scope_requires_projection_repair() ->
             }
         ],
     }
-    relation = todo_gate_relation(gate, agent_item)
+    relation = todo_gate_relations([gate], [agent_item])[0][0]
     assert relation["source"] == "unblocks_todo_id+decision_scope", relation
     assert relation["state"] == "projection_repair_required", relation
     assert relation["reason"] == (

@@ -276,6 +276,7 @@ export function createEffectRuntimeHandlers(
     ["todo.succession.project", lazyHandler(() => import("./todos/succession.ts"), ({projectTodoSuccession}) => projectTodoSuccession)],
     ["todo.work_counts.project", lazyHandler(() => import("./todos/summary_lanes.ts"), ({projectLegacyTodoWorkCounts}) => projectLegacyTodoWorkCounts)],
     ["projection.envelope.seal", lazyHandler(() => import("./projection_envelope.ts"), ({sealProjectionEnvelope}) => sealProjectionEnvelope)],
+    ["todo.global_readiness.project", lazyHandler(() => import("./todos/decision_scope.ts"), ({projectGlobalTodoReadiness}) => projectGlobalTodoReadiness)],
     ["todo.decision_scope.evaluate", lazyHandler(() => import("./todos/decision_scope.ts"), ({evaluateDecisionScope}) => evaluateDecisionScope)],
     ["todo.user_completion.plan", lazyHandler(() => import("./todos/user_completion.ts"), ({evaluateUserCompletion}) => evaluateUserCompletion)],
     ["agent.capability_gate.evaluate", lazyHandler(() => import("./agents/capability_gate.ts"), ({evaluateCapabilityGate}) => evaluateCapabilityGate)],

@@ -14,7 +14,7 @@
     normalize_todo_id,
     normalize_todo_required_decision_scopes,
 )
-from .user_gate import is_user_gate_todo_item
+from .user_gate import is_user_gate_todo_item, open_user_gate_todo_items
 from .todo_semantics import todo_projection_sort_key, todo_item_task_class, todo_item_has_removed_continuation_policy
 
 DECISION_SCOPE_CONSISTENCY_SCHEMA_VERSION = "required_decision_scope_consistency_v0"
@@ -293,13 +293,6 @@ def build_required_decision_scope_repair_hint(
     return result
 
 
-def todo_gate_relation(gate: dict[str, Any], agent_item: dict[str, Any]) -> dict[str, Any] | None:
-    return _optional_relation(
-        "relation",
-        _evaluate("relation", gate=_facts(gate), item=_facts(agent_item)),
-    )
-
-
 def todo_gate_relations(gates: list[dict[str, Any]], items: list[dict[str, Any]]) -> list[list[dict[str, Any] | None]]:
     """Evaluate a consumer's candidate set in one RPC, retaining positional identity."""
     if not gates or not items:
@@ -356,3 +349,60 @@ def facts(item: dict[str, Any]) -> dict[str, Any]:
     if not isinstance(projections, list) or len(projections) != len(items):
         raise TypeError("invalid typed gate scope projection cardinality")
     return [_required_projection("gate_scope", row, schema_version="todo_gate_scope_projection_v0") for row in projections]
+
+
+def _readiness_facts(item: dict[str, Any]) -> dict[str, Any]:
+    return {
+        **_facts(item),
+        "identity_text": str(item.get("todo_id") or ""),
+        "status_text": str(item.get("status") or ""),
+        "target_text": str(item.get("unblocks_todo_id") or ""),
+    }
+
+
+def _global_readiness_request(payload: dict[str, Any]) -> dict[str, Any]:
+    """Bound transport to candidate containers; TS alone selects their list keys."""
+    bounded = {
+        key: payload.get(key)
+        for key in ("selected_todo", "agent_lane_next_action", "normal_delivery_allowed")
+    }
+    decoded = {
+        f"/{key}": _readiness_facts(value)
+        for key, value in bounded.items()
+        if isinstance(value, dict)
+    }
+    for container_key in ("agent_todo_summary", "agent_scope_frontier"):
+        container = payload.get(container_key)
+        lists = {
+            key: value for key, value in container.items() if isinstance(value, list)
+        } if isinstance(container, dict) else {}
+        bounded[container_key] = lists
+        for key, values in lists.items():
+            for index, item in enumerate(values):
+                if isinstance(item, dict):
+                    decoded[f"/{container_key}/{key}/{index}"] = _readiness_facts(item)
+    return {
+        "schema_version": "global_todo_readiness_request_v0",
+        "payload": bounded,
+        "decoded": decoded,
+        "gates": [
+            _readiness_facts(gate)
+            for gate in open_user_gate_todo_items(payload.get("user_todo_summary"))
+        ],
+    }
+
+
+def project_global_todo_readiness(payload: dict[str, Any]) -> dict[str, Any]:
+    """Decode persisted fields and reuse the shared Python gate selector."""
+    result = effect_runtime_result(
+        "todo.global_readiness.project", _global_readiness_request(payload),
+    )
+    projection = _required_projection(
+        "global_readiness", result, schema_version="global_todo_readiness_v0",
+    )
+    for key in ("warnings", "candidates"):
+        if not isinstance(projection.get(key), list) or any(
+            not isinstance(row, dict) for row in projection[key]
+        ):
+            raise TypeError("invalid typed decision scope global readiness rows")
+    return projection